AI Moderation: Architected an LLM-driven detection framework that automates 2,500+ daily moderation tasks with 97% precision, significantly reducing manual overhead.

Infrastructure Optimization: Engineered a high-efficiency feature extraction pipeline that slashed LLM token costs by 66% while maintaining real-time defense coverage.

Multimodal Safety: Deployed specialized CLIP based classifiers to identify high-risk content proactively, preventing harmful exposure across millions of Stories.

Scalable Data Engineering: Built robust feature datastores using Apache Beam (Dataflow) to aggregate graph embeddings and user-behavior signals for content-agnostic threat detection.

Working on the API's for the privacy team to redact customer data from Alexa before reaching annotators. Worked on various projects like improving audit reports, onboarding new customers, adding throttling engine, onboarding new models to the team

Worked in the partner center billing and payments team over API, Data and UI platforms

Teaching Assistant for Natural Language Processing and Machine Learning for Fall 2017